Core temperature modelling and monitoring of lithium-ion battery in the presence of sensor bias

Abstract

Lithium-ion battery is now considered as an enabling technology for modern civilization and sustainability, initiating wireless revolution and efficient energy storage. In spite of the significant progress made in materials and manufacturing, the growing awareness on the operational safety and reliability requires more efficient battery management systems. Core temperature, an underlying variable for battery management, is unfortunately unmeasurable and has to be online estimated via other measurable variables. The core temperature monitoring becomes even more challenging when the measurable variables encounter sensor bias as well as model inaccuracy and sensor noise. To this end, this paper improves the thermal model accuracy by introducing a radiation term into the conventional linear lumped model. The unknown parameters of the new nonlinear model are identified based on multi-objective optimization, of which the results confirm the superiority of the proposed nonlinear model. The sensor bias is treated as an extended state to be estimated together with other states. An extended unscented Kalman filter is accordingly developed to handle the nonlinearity, measurement noise and sensor bias. Both simulation and experimental results are given to demonstrate the efficacy of the proposed method, showing that the core temperature can be accurately estimated in spite of the sensor biases in the surface temperature or electric current measurements.